AI SIGNAL
Curated frontier AI & tech news, refreshed every four hours. Aggregated from Hacker News, GitHub trending, and Hugging Face / Replicate releases.
[?] how ranking works
- 501+ exceptional
- 100–500 strong
- 51–99 mid
- 30–50 fresh
- <30 quiet
- live — currently trending
- cooling — losing momentum (6–18h)
- frozen — no longer trending; ranking decays over time
Composite ranking: trend velocity + log(raw signal) − staleness penalty. The bar shows relative strength within the current batch, higher = stronger overall position.
Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s
Strata runs the Qwen3.8-Flash-Next 125B model on consumer hardware like RTX 4090s, supporting NVIDIA/AMD cards with 12GB+ VRAM and offering speeds up to 100T/s on 24GB cards.
OpenAI safety leader quits, warning AI company's culture is 'broken'
OpenAI's safety leader has quit, warning the company's culture is 'broken'.
Blindsight (Watts Novel)
Blindsight is a 2006 hard sci-fi novel by Peter Watts about a transhuman crew investigating a comet transmitting an alien signal, exploring consciousness and first contact themes.
In Ukraine, distributed renewables foil Russia's assaults
Distributed solar and renewables in Ukraine provide critical water and energy resilience against Russian attacks, making them harder to target than centralized fossil fuel plants.
Rejection Sensitivity in Gifted and Twice-Exceptional Children
Rejection sensitivity in gifted and twice-exceptional children stems from intense nervous systems and identity fusion, manifesting as RSD, sensory sensitivity, or overexcitabilities, and is not manipulation or bad parenting.
Why don't more developers “use the platform”?
Developers avoid native APIs due to historical browser gaps, npm convenience, better documentation, and the fun of building custom solutions like modals, despite performance benefits.
Show HN: AI search for every photo and every frame of video on macOS
SCM is a local-first macOS app that uses AI to search photos and video frames via natural language, OCR, and dialogue, with features like scene segmentation and self-maintaining libraries.
We're going to need default hard budget caps on pretty much everything
Simon Willison advocates for default hard budget caps on pay-by-usage services to prevent runaway costs from coding agents, noting AWS and Google Cloud have recently launched similar features.
Big Balls Now Exposed to Serious Criminal Charges in at Least Six States
Scott Kupor praised 'Big Balls' as a model federal employee despite his history of leaking secrets, running a CDN for a cybercrime group, and being named in a Social Security whistleblower complaint.
Apple's "Clean Design" Is Stupidity When It Comes to Hiding Fire Extinguishers
Hiding fire extinguishers behind unmarked panels violates OSHA and NFPA safety regulations, making them inaccessible during emergencies.
VGHF Digital Archive passes 5000 magazines. Here's what's next
The Video Game History Foundation has added 5000+ magazines, totaling 600,000 pages, and now supports Japanese-language magazines like Neo Geo Freak with advanced OCR and English article indexing.
Religious scholars met with Anthropic
Religious scholars met with Anthropic to discuss AI morals, with Christopher Olah stating he is genuinely uncertain if AI models are conscious.
Federal judge calls Flock 'indiscriminate mass surveillance'
A federal judge ruled a Flock Safety search unconstitutional, calling it indiscriminate mass surveillance, and ordered evidence suppressed.
Emitting metadata early makes building/checking Rust up to twice as fast
Headstart patches rustc and cargo to emit early metadata, allowing dependent crates to start compiling before their dependencies finish type-checking, speeding up builds by up to 54%.
We're working on a new RuneScape MMO
Jagex is developing RS4, a new RuneScape MMORPG built in Unreal Engine set in Gielinor's Ashenfall region, designed to sit alongside existing games rather than replace them.
UK Government Body Kept Files on People Criticizing Prevent Program
The UK Home Office's Prevent unit tracked 77 online critics of the program on X and Reddit, storing their posts as 'open-source complaints' despite the government's initial attempt to keep the data secret.
Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥
Firecrawl is an open-source API for web scraping and data extraction, offering search, crawl, and agent features with LLM-ready output and high reliability.
π RuView turns commodity WiFi signals into real-time spatial intelligence, vital sign monitoring, and presence detection — all without a single pixel of video.
RuView uses ESP32 WiFi sensors to detect people, breathing, and movement through walls without cameras or internet, integrating with Home Assistant, Apple Home, Google Home, and Alexa.
An advanced guide which might benefit you a lot 🎉 . 韩先凯的人生进阶指南 人生进阶指南 离谱的人生 人生进阶 AI学习 AI指南 韩先凯的AI学习指南 英语学习指南/英语学习教程/英语学习/学英语
GitHub project byoungd/up with 67,096 stars.
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Claude-Mem is a persistent memory compression system for Claude Code that automatically captures tool usage and generates semantic summaries to maintain context across sessions.
Turn any idea, plan, or codebase into a beautiful interactive diagram. An agent skill for Claude Code, Codex, and more.
Archify is a Node.js tool that converts codebase JSON into interactive, shareable system architecture diagrams for AI coding agents like Cursor and Claude.
Spotify, native and fast. One lightweight Rust app for your whole library, local playback, and Spotify Connect on Linux, macOS, and Windows.
Spotifast is a fast, native Spotify client written in Rust that uses 100-250 MB RAM, starts in under a second, and requires Spotify Premium for playback, with features like device control, library browsing, and playlist editing.
高性价比人生指南: 长寿防病、急救、省钱理财、法律红线、失业与工伤、医保社保、恋爱婚育、怀孕育儿、创业与做平台合规、出国与技能。每条写明成本、收益、证据等级和原始出处，只引期刊论文与官方文件。
This GitHub repository is a practical guide covering 32 topics including health, money, law, relationships, and safety, answering specific questions like how to avoid early death, manage chronic illness, and handle legal risks.
The agent that grows with you
Hermes Agent is a self-improving AI agent with built-in learning loops, memory, and multi-platform support (Telegram, CLI, etc.), running on any model via Nous Portal or OpenRouter, and installable via a simple script.
Python tool for converting files and office documents to Markdown.
MarkItDown is a Microsoft Python utility converting files like PDF, Word, and images to Markdown for LLMs, supporting optional dependencies and plugins.
Give your agent CAD superpowers.
A library of agent skills for generating, inspecting, and fabricating CAD models, robot descriptions, and 3D prints from text or images.
Experiential is the open source, zero markup gateway for BYOK, self-hosted and 1000+ marketplace models. It learns from your traffic to cut costs, recommend better models, and train a specialized model you own.
Experiential is an open source gateway and router for agent workflows that unifies hosted, BYOK, and local models through an OpenAI-compatible API, allowing users to control access, budgets, and optimize models using production traffic.
AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI
Pi is a self-extensible coding agent harness with packages for AI, agent core, and a CLI, plus strict supply-chain hardening and containerization options.
Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
Free Claude Code is an open-source tool aggregating 49 free AI coding providers and 9 agents like Claude Code and Cline, offering a unified UI, terminal optimizations, and support for multiple platforms.
25 MB lightweight cross-platform database client for 100+ databases, including MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, SQL Server, and Dameng. Built-in AI, MCP Server, CLI, desktop and Docker. | 轻量级跨平台数据库管理工具，支持 MySQL、PostgreSQL、SQLite、Redis、MongoDB、达梦等 100+ 数据库，提供桌面端、Docker、CLI、内置 AI 助手和 MCP。
GitHub project t8y2/dbx with 24,481 stars.
Make videos on the computer you already own. FreeVideo runs MiniMax H3 in as little as 8 GB of VRAM and 16 GB of RAM, and adapts its acceleration path to your hardware.
FreeVideo is a local inference engine for MiniMax H3 on consumer GPUs, running on as little as 8GB VRAM and 16GB RAM, with hardware-adaptive execution and ComfyUI integration.
A super fast Graph Database uses GraphBLAS under the hood for its sparse adjacency matrix graph representation. Our goal is to provide the best Knowledge Graph for LLM (GraphRAG).
FalkorDB is an ultra-fast, multi-tenant graph database built in Rust, designed to power generative AI, agent memory, and fraud detection with high performance.
手绘风格编号画廊与双语提示词 Skill
A prompt library for AI image generation offering 279 hand-drawn styles, 122 layouts, and 36 color themes to create social media graphics, comics, and posters.
A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
This GitHub page promotes Kimi K2.7 Code, an open-source coding model, and lists sponsored API relay services like PackyCode, AIGoCode, and TeamoRouter with specific discount codes and benefits for users.
Node.js JavaScript runtime ✨🐢🚀✨
Node.js is an open-source JavaScript runtime environment with Current, LTS, and Nightly release types, governed by the OpenJS Foundation and TSC, and available for download at nodejs.org.
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
Colibrì is a pure C engine running the massive 744B GLM-5.2 MoE model on consumer hardware by streaming experts from disk, featuring MLA attention, int8 MTP speculation, and 9.9GB resident RAM usage.
**Code on GitHub** · **Demo** · **Benchmark: razvilka** · **Colab quickstart** · **vLLM server** **FRIDA-Decisions** makes structured decisions over Russian text in a single encoder pass: pick one...
FRIDA-Decisions is a Russian NLP model for structured text decisions (choice, ranking, yes/no) using a single encoder pass, achieving 0.893 on the razvilka benchmark with 28-34ms latency on an RTX 5060 Ti.
A modern GUI client based on Tauri, designed to run in Windows, macOS and Linux for tailored proxy experience
GitHub project clash-verge-rev/clash-verge-rev with 149,069 stars.
오철칙(Iron Laws) - AI가 만든 코드의 보안·구조·유지보수 문제를 짚어 주는 소스코드 점검 CLI
A security and quality linting tool for AI-generated code, based on government audit standards, that checks for vulnerabilities, structural issues, and type safety while generating fix prompts for AI.
12 电商出图 Agent Skills for Claude Code / Codex / Cursor · GPT Image 2.5 (Flare/Sunburst) · Amazon 白底主图, A+, 场景图, 模特图, 小红书封面, 淘宝详情页
A GitHub repo providing 12 modular Agent Skills for generating ecommerce product images across platforms like Amazon, Taobao, and TikTok Shop, featuring product identity locking, platform-specific specs, and model routing (Flare/Sunburst).
MiniMax H3 (video + audio) on TensorFold 0.6.5 with an H3 family, Mac Studio M5 Ultra: 5 s 864x480 clip in ~50 s, 40 s at 768x448 — the fastest H3 we have measured on Apple Silicon; int8 tensor-unit kernels + Turbo adapter; one-shot + GHCR carrier
This repo benchmarks TensorFold's MiniMax H3 implementation on Mac Studio M5 Ultra, showing it generates 864x480 clips in ~50s using Turbo adapter and int8 kernels, outperforming 20-step engines by 4x speed.
SC117/Swift-1.5-Qwen3.8-Flash-Next-GSQ-RCO-abliterated-GGUF
This is a GGUF model page providing installation and usage instructions for libraries like llama.cpp, vLLM, Ollama, and Pi, including specific commands for running the model locally.
Qwen image 2.1 Enhancement Nodes
Adds reference image strength and prompt phrase weighting nodes for Qwen Image 2.1 in ComfyUI, allowing users to adjust attention priority for specific images and prompt parts.
Run with https://llama.app ```bash llama serve -hf ggml-org/Laya-GGUF ``` This is a decision model, to be used via `/v1/systemone` API. See https://github.com/ggml-org/llama.cpp/pull/29818
Laya-GGUF is a decision model hosted on Hugging Face, available in BF16 format for use with llama.cpp, Ollama, and other local apps via specific commands like llama serve -hf ggml-org/Laya-GGUF:BF16.
Tensors and Dynamic neural networks in Python with strong GPU acceleration
PyTorch is a Python library for tensor computation and deep learning with GPU acceleration, featuring a tape-based autograd system and components like torch.nn and torch.jit.
Visual Studio Code
Microsoft's open-source Visual Studio Code repository (Code- OSS) is a free, extensible code editor available under the MIT license, with monthly updates and a rich community contribution model.
Agent skill for designing, mocking, building and auditing terminal UIs
An agent skill for designing, mocking, and auditing terminal UIs with evidence, generating linted frames, HTML galleries, and code starters for frameworks like Textual and Ink.
Multi-chain MEME radar with deterministic evidence, ARC Mainnet monitoring, and isolated execution boundaries
An open-source, local-first meme discovery system with four paths: Monitor, Frontend, Intelligence, and Execution, featuring safe-by-default security and ARC Mainnet support.
Paper (arXiv) · Project page This repository holds the **LoRA adapter** of a 4-step (4 NFE) student distilled from MiniMax-H3 with **Projected Distribution Matching Distillation (PDMD)**.
A 4-step LoRA adapter for MiniMax-H3 using Projected Distribution Matching Distillation, containing 312 LoRA pairs with rank 128 and alpha 128, enabling 4 denoising steps.
tmux source code
tmux is a terminal multiplexer that allows you to run multiple terminal sessions inside a single terminal window, useful for managing long-running processes and multitasking.
Signal briefs — our take on the top picks
- HNBlindsight (Watts Novel)
- HNRejection Sensitivity in Gifted and Twice-Exceptional Children
- HNRun Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s
- HNIn Ukraine, distributed renewables foil Russia's assaults
- HNWhy don't more developers “use the platform”?
- HNBig Balls Now Exposed to Serious Criminal Charges in at Least Six States
- HNOpenAI safety leader quits, warning AI company's culture is 'broken'
- HNWe're going to need default hard budget caps on pretty much everything
- HNGetting the most out of Opus 5.5 in Claude and Claude Code
- HNHole Punch: Sling your spaceship around gravitational fields
- HNFederal judge calls Flock 'indiscriminate mass surveillance'
- HNCity building games have a Soul Problem pt.2